Jun 6, 2024 · 1h 5m · in-depth
Developing technical taste: A guide for next-gen engineers | Sam Schillace (Microsoft, Google Docs)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this in-depth conversation, Microsoft Deputy CTO and Google Docs creator Sam Schillace shares hard-won principles on market timing, technical taste, and building high-performing engineering cultures across major technology shifts. Drawing on thirty years of software leadership, Schillace provides tactical frameworks for navigating generative AI, maintaining radical optimism, and mastering the evolving craft of software engineering.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Brett holds 13.2% of the talking time here. How this is scored →
Brett as informed peer 3.4 Guest teaching 5.5 Guest disagreement 1.4 Brett pushing back 0.5
speaking balance: gold is Brett, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 59:20 Dismantling the Luxe valet comparison
Sam rejects Brett's comparison to startup fads like Luxe, bluntly noting that those businesses lacked positive unit economics or scale effects and were merely pumped with venture cash.
Hardest push from Brett ▶ 22:50 Challenging the premise of total Google Docs disruptionBrett pushes back against the clean narrative of disruptive innovation by pointing out that Microsoft Office was not wiped out like Blockbuster and remains massive.
Biggest teaching moment ▶ 16:25 Masterclass on the distributed architecture of Google DocsSam walks through the intricate reality of building Writely, explaining why differing browser DOM trees broke text diffing and necessitated building early operational transforms from scratch.
Brett holds their own ▶ 58:46 Brett citing historical bubble failures to question aggressive forecastingBrett demonstrates market pattern recognition by citing on-demand valet blowups like Luxe to challenge overconfident technological predictions.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Brett as informed peer | Guest teaching | Guest disagreement | Brett pushing back | Why |
|---|---|---|---|---|---|---|
| Introducing Sam Schillace and Episode Overview | 3 | 4 | 1 | 0 | Brett introduces Sam's storied engineering career across Google Docs, Box, and Microsoft, and asks a broad scene-setting question about what decades of building software teach about market timing. Sam explains that working on uncomfortable, unsolved problems is where asymmetric upside lies. | |
| Calculated Optimism and the Risks of Market Timing | 4 | 5 | 2 | 2 | Brett probes whether being too early is an overrated risk. Sam nuances the point by arguing that failing early leaves room to course-correct, whereas failing late is fatal, drawing an analogy to solar energy learning curves. | |
| Developing Technical Taste and Recognizing Engineering Quality | 3 | 6 | 1 | 1 | Brett asks Sam to define 'technical taste'. Sam educates the host with concrete heuristics and vivid analogies, comparing well-run software teams to chasing a toddler down a hill and citing SpaceX's disciplined engineering. | |
| High-Performing Teams: 'What If' vs. 'Why Not' Mindsets | 4 | 5 | 1 | 0 | Brett observes how tech company risks shifted between technical and market risk over the past decade. Sam introduces the cultural dynamic of 'what if' exploratory teams versus 'why not' cynical teams. | |
| Generating Insights Through Rapid, Low-Cost Experiments | 2 | 5 | 0 | 0 | Brett asks how to evaluate the success and progress of experimental prototyping teams. Sam explains the discipline of cheap, rapid experiments and fast proof points rather than long theoretical builds. | |
| The Architecture and Evolution of Google Docs | 2 | 7 | 1 | 0 | Brett asks about the origin of Google Docs. Sam delivers a detailed engineering breakdown of Writely, the struggle with browser tree diffing, early operational transform, moving backend systems to Java and Spanner, and rewriting canvas renderers. | |
| Feature Trade-offs, User Demands, and Enterprise Bloat | 3 | 5 | 2 | 0 | Sam explains how refusing feature bloat allowed Google Docs to beat Word, though student demand forced adding word counts. He candidly critiques the current bloat and degradation of Google Docs and Gmail. | |
| Disrupting Incumbents by Changing the Terms of Engagement | 5 | 6 | 2 | 2 | Brett connects the Docs narrative to Christensen's disruptive innovation framework and questions whether it really disrupted Office. Sam reframes the dynamic around changing terms of engagement and enterprise trust barriers. | |
| Engineering Pressures: Consumer Friction vs. Enterprise Contracts | 4 | 6 | 1 | 0 | Brett asks about differences in engineering culture between consumer and enterprise software. Sam clearly articulates how consumer software requires relentless user friction reduction while enterprise software is anchored in contractual trust and CIO SLAs. | |
| Why Engineers Must Resist Cynicism and Embrace Possibility | 3 | 6 | 2 | 1 | Brett asks why engineering teams continue to grow despite massive layers of abstraction. Sam explains Parkinson's law in tech and details how generative tools will inflate ambitions rather than solely shrinking team headcount. | |
| Generative AI and the Emergence of On-Demand Fluid Software | 3 | 6 | 1 | 0 | Sam outlines his thesis on zero-cost pixel creation, contrasting handcrafted static software with on-demand, fluid software and questioning why modern computers still imitate typewriter paradigms. | |
| Capturing Value in AI: Personal Data, Trust, and Scaffolding | 4 | 6 | 1 | 0 | Brett asks who captures value in this shift. Sam outlines why trusted personal data stores and deterministic software scaffolding around stochastic models form durable defensive moats. | |
| The Cognitive Surplus Era and Reimagining Organizational Hierarchy | 4 | 6 | 1 | 0 | Brett asks how historical cycles inform current judgment. Sam draws the parallel between the physical energy surplus of the Industrial Revolution and the cognitive surplus of modern AI, projecting flattened corporate hierarchies. | |
| Comparing Early Cloud Skepticism to Current LLM Doubts | 3 | 5 | 2 | 0 | Sam compares today's LLM skepticism with early skepticism of cloud storage and Google Docs, recounting his offer to trade and wipe laptops with doubters. | |
| Career Guidance for Young Engineers: Cultivating Play and Curiosity | 3 | 5 | 2 | 0 | Brett asks what advice Sam would give ambitious 22-year-old computer science graduates. Sam pushes back against overly curated academic perfectionism, urging new grads to make messes, play, and treat LLMs as tools rather than oracles. | |
| Core Engineering Excellence: Lucidity, Simplification, and Adaptability | 3 | 5 | 1 | 0 | Brett inquires about enduring traits of elite engineers. Sam highlights clarity of thought and the ability to simplify complex systems, explaining that programming languages are ephemeral. | |
| Prototyping Workflows: Vector Memory, Mermaid Graphs, and Trapdoors | 3 | 7 | 1 | 0 | Brett asks for concrete examples of modern AI prototyping workflows. Sam reveals how his team builds rag-based bots with vector memory, renders Mermaid flowcharts on the fly, and conceptualizes tasks as cryptographic trapdoors. | |
| Strategic Decision-Making for Incumbent Software Businesses | 4 | 5 | 1 | 0 | Brett asks how incumbents like TurboTax or Workday should respond to generative AI. Sam advises running thought experiments where all current AI blockers are completely solved and reassessing the core customer value proposition. | |
| Assessing Technological Bets: Unit Economics vs. Data Scale | 5 | 6 | 3 | 3 | Brett raises failed VC-backed bets like Luxe valet services to challenge aggressive forecasts. Sam briskly distinguishes between flawed business models lacking positive unit economics and legitimate tech scale curves like autonomous driving and diagnostic data. | |
| AI Integration in Medicine: De-biasing and Cognitive Leverage | 3 | 5 | 1 | 1 | Sam explores AI's role in medicine as a de-biasing tool for physicians and closes the interview reflecting on lifelong curiosity and building beloved tools like Google Docs. |
Statements from this episode (22)
Schillace: Waiting for tech to get easier guarantees too much competition
“The mistake people make the most is, you know, they'll be like, well, that's a cool idea, but it's really hard. Let's wait three years till it's easier. When it's easier, there'll be a million people doing it and there's path dependencies in there and stuff li…”
Schillace: Over one billion people use Google Docs today
“There's like more than a billion people use gdocs today.”
Schillace: Failing early is better than failing late in technology
“Failing on the early side, you at least have a chance to, like, course correct, right? Because if you don't spend all your resources and burn out completely, you can keep going, and maybe the world catches up to you, or maybe you make the world catch up to you…”
Schillace: There is no prize for being pessimistic and right in tech
“There's not really much of a prize for being pessimistic and right about these things. Ok, you get a ribbon. Who cares? The prize is for being optimistic and right.”
Schillace: You cannot short broad technological revolutions
“You can't go short on, like, technical, technological revolution, right? You can't, like, short solar or some crazy shit like that. It tends to be wrong. We tend to figure stuff out over time. You know, at best, you're right for a little while, and eventually …”
Schillace: Successful software projects feel slightly out of control
“Most of the successful software projects I've seen Have the same feeling of sort of slightly out of control, but not too much of a disaster as chasing a three year old down a slight incline in a park.”
Schillace: Microsoft's experimental AI team explores agent mechanics instead of shipping products
“I've got a team right now at Microsoft that's This sort of experimental prototyping team. It's just the whole thing is like a jazz ensemble. It's really wild. We don't really ship products. We're trying to explore the space and understand some deep things abou…”
Schillace: Writely ran on only three servers before Google acquisition
“We had none of the backend support that you have now in the cloud. Like we literally had three servers in Texas. And so scaling was starting to get to be a problem when we went to Google.”
Schillace: Google Docs has become a complicated, bloated version of Microsoft Office
“To be honest, like, I'm a little bit sad at the state of Gdocs these days because I think they're losing sight of the idea that, you know, they're kind of turning it into Office and like making it very complicated to use again. And I find lots of things broken…”
Schillace: Microsoft cloud tools must preserve desktop compatibility, unlike Google Docs
“They're trying to keep the cloud very strongly competitive roundtrip, or very strongly compatible, you know, roundtrip compatible with the desktop. That's a design point Google does not care about, but Microsoft does care about.”
Schillace: Google Workspace and Microsoft Office are reaching market stasis
“And I think it's hard for either one company to fully become what the other one is and take, take the whole market. So I think we're getting close to stasis on this actually.”
Schillace: Enterprise software de-emphasizes UX once it clears the CIO's checklist
“There's less of an emphasis on end user experience, I think in the enterprise space, because once you're kind of over the line where the CIO says, okay, that's on the checklist of things that your software does, anything above that doesn't really matter very m…”
Schillace: Google spends billions annually on Google Maps
“Google spends billions of dollars a year on maps.”
Schillace: AI tooling will expand engineering ambitions rather than shrink team sizes
“Programmers are definitely going to get more effective. We're going to get more agentic programming tools. So you're going to get higher and higher leverage, but I don't think it's going to make the team smaller. I think it's just going to make the ambitions b…”
Schillace: AI will make creating pixels free, radically altering software applications
“I have the suspicion that the nature of applications is going to change radically in the next five to 10 years. And so I think, you know, this idea that just in the same way that the internet made the distribution of information essentially free, AI, generativ…”
Schillace: Many businesses will be disrupted because their models rely on pixel expense
“I think there are lots of businesses out there that don't fully understand that they're predicated on the expense of creating pixels. And so similar disruption is in the works.”
Schillace: Durable AI value will accrue to trusted personal data holders
“So that's kind of where that durable value will wind up being, I think, over time. It's kind of the holders of the data and the companies that we can trust to, you know, hold that data and trust to give us the right answers are probably where their value is.”
Schillace: Engineers should think with AI models but plan with code
“The easy way to understand this is when I say you should think with the model, but plan with code because the models are, they're stochastic and fuzzy. And so if you want something reliable, you should build kind of a scaffold around it in code, something exec…”
Schillace: AI chatbots should be managed and edited like documents
“We have this idea that these chatbots should be treated like documents. I say bots or docs a lot. So meaning that they have individual names and you can edit all the pieces of them and you can save them and share them and delete them and all this copy them and…”
Schillace: High-value LLM applications work like a cryptographic trapdoor
“The things that work well with LLMs have this shape almost like a cryptographic trapdoor. There are like things where it's started work for the human to create the work, but easy for the human to validate the work. Those are the things that tend to have high v…”
Schillace: Incumbents must assume AI blockers get solved and reassess value propositions
“Whatever it is that's in the way of this disrupting your business. Now, like assume that that gets solved. What's your business look like after that gets solved? And then what do you do? What is the thing that your business is about? What value are you deliver…”
Schillace: Physicians paired with AI diagnostic tools perform worse than AI alone
“There's a study I saw a while ago that compared like an AI based system, AI only physician only and AI plus physician An AI plus physician did worse than AI only. The physician kind of dragged the AI down because they didn't trust it.”